Custom model hosting

Your model, operated as part of CarHub.

Bring a model that is specific to your vehicles, markets or workflow. We package it, operate it and expose it through the same reliable platform your product already uses.

Vehicle wireframe connected to GPU compute hardware

GPU compute

Choose the hardware your model needs.

Prices are in USD. Per-second pricing keeps a pilot lean and lets production capacity scale with demand.

HARDWAREPER SECOND
  • NVIDIA B300$0.002958
  • NVIDIA B200$0.002604
  • NVIDIA H200 SXM$0.0018915
  • NVIDIA H100 SXM5$0.0016455
  • NVIDIA RTX PRO 6000$0.001263
  • NVIDIA A100, 80 GB$0.001041
  • NVIDIA A100, 40 GB$0.0008745
  • NVIDIA L40S$0.000813
  • NVIDIA A10$0.000459
  • NVIDIA L4$0.000333
  • NVIDIA T4$0.000246

Supporting resources

CPU, memory and storage pricing.

CPU and memory are billed per second. Persistent volumes are billed per GiB each month.

  • CPU

    Physical core (2 vCPU)

    PRICE

    $0.00001965

    per core / sec

  • Memory

    Allocated memory

    PRICE

    $0.00000333

    per GiB / sec

  • Volumes

    Persistent model weights and data

    PRICE

    $0.135

    per GiB / month

Have a model ready to run?

Tell us about the model, its inputs and the workload you expect. We will help select the right hardware and map out the deployment.

Talk to us about model hosting